Remember this broken Giraffatitan dorsal vertebra, which Janensch figured in 1950?

It is not only cracked in half, anteroposteriorly, it’s also unfused.

Here’s a better view of the broken face, more clearly showing that the neural canal is (a) much taller than wide – unlike all vertebrate spinal cords – and (b) almost entirely situated ventral to the neurocentral joint, getting close to the condition in the perverted Camarasaurus figured by Marsh.

Here’s a dorsal view, anterior to the top, with Mike’s distal forelimbs for scale.

Left lateral view.

Right lateral view – note the subtle asymmetries in the pneumatic foramen/camera. A little of that might be taphonomic distortion but I think much of it is real (and expected, most pneumatic systems produce asymmetries).

And postero-dorsal view, really showing the weird neural canal to good advantage. In this photo and in the pure dorsal view, you can see that the two platforms for the “neural arch” – which, as in the aforementioned Camarasaurus, is neither neural nor an arch – converge so closely as to leave only a paper-thin gap.

A few points arise. As explained in this post, it makes more sense to talk about the neurocentral joint migrating up or down relative to the neural canal, which is right where it always is, just dorsal to the articular faces of the centrum.

So far, in verts I’ve seen with “offset” neurocentral joints, the joint tends to migrate dorsally in dorsal vertebrae, putting the canal inside the developmental domain of the centrum (which now includes a partial or total arch in an architectural sense, even though the chunk of bone we normally call the neural arch develops as a separate bit) – as shown in the first post in this series. In sacral and caudal vertebrae, the situation is usually reversed, with the joint shifted down into what would normally be the centrum, and the canal then mostly or completely surrounded by the arch – as shown in the second post in the series. This post then doesn’t really add any new concepts, just a new example.

Crucially, we can only study this in the vertebrae of juveniles and subadults, because once the neurocentral joints are fused and remodeled, we usually can’t tell where the old joint surface was. So it’s like cervicodorsal and caudal dorsal pneumatic hiatuses, in that the feature of interest only exists for part of the ontogeny of the animal, and our sample size is therefore inherently limited. Not necessarily limited by material – most museums I’ve visited have a fair amount of juvenile and subadult material in the collections – but limited in published visibility, in that for many sauropods only the largest and most complete specimens have been monographically described.

So once again, the answer is simply to visit collections, look at lots of fossils, and stay alert for weird stuff – happily, a route that is open to everyone with a legitimate research interest.


  • Janensch, W. 1950. Die Wirbelsaule von Brachiosaurus brancai. Palaeontographica (Suppl. 7) 3:27-93.

This is the second post in the “bird neural canals are weird” series (intro post here), and it covers the first of five expansions of the spinal cord or meninges in the lumbosacral regions of birds.

The lumbosacral expansion of the spinal cord is not unique to birds and doesn’t require any special explanation. As noted in the slide, all limbed tetrapods and some fishes with sensitive fins have adjacent segments of the spinal cord correspondingly expanded. These expansions house the extra afferent neurons needed to collect sensory inputs from the limbs, the extra efferent neurons needed to provide motor control to the limbs, and the extra interneurons needed for sensory and motor integration (including reflex arcs) – ‘extra’ here meaning ‘more than are required for non-limb neck, trunk, and tail segments’.

Humans have these, too, in our lower cervical vertebrae to run our forelimbs, and in our lower thoracic vertebrae to run our hindlimbs. Recall that the segmental anatomy of the adult human spinal cord corresponds increasingly poorly to the vertebrae the farther we are from the head because of our child-sized spinal cords (see this post for more).

So if the lumbosacral expansion is present in all tetrapods with hindlimbs, why bring it up? My goal is to develop a set of criteria to distinguish the various spinal and meningeal specializations in birds, in part because it’s an interesting challenge in its own right, and in part because doing so may help illuminate some unusual features in sauropods and other non-avian dinosaurs. If we want to be able to detect whether, say, a glycogen body is present, we need to know how to tell the impression left by a glycogen body from the more generalized lumbosacral expansion present in all limbed tetrapods. The key characteristics of the lumbosacral expansion are that the cord (and hence the canal) expands and contracts gradually, over many segments, and that the expansion is in all directions, radially, and not biased dorsoventrally or mediolaterally.

Numbering reflects spinal nerve count – 8 cervical, 12 thoracic, 5 lumbar, and 5 sacral spinal spinal nerves. Cervical expansion for the forelimbs is roughly C5-T1, and lumbosacral expansion for hindlimbs is L2-S3. Gray (1918 image 665).

The one way in which the lumbosacral expansion of birds is weird, at least compared to mammals, is that the magnitude of the change is so great in hindlimb-dominant flightless birds like the ostrich. Here’s a graph from Gray’s Anatomy showing the cross-sectional area of the human spinal cord in square mm, with the head on the left. Note that the swellings for the limbs bump up the cross-sectional area by a quarter to a third, relative to adjacent non-limb areas.

Streeter (1904: fig. 4)

Here’s the same diagram for an ostrich, again in square mm, again with the head to the left. The lines here are a little different – the “substantia grisea” is the gray matter (mostly neuron cell bodies), and the white matter (axons, mostly myelinated) is divided into the large ventrolateral funiculi (descending motor, ascending pain, temperature, and unconscious proprioception) and the much smaller dorsal funiculi (ascending touch and conscious proprioception). Here the lumbosacral expansion maxes out at more than double the cross-sectional area of the cord in the inter-limb torso segments – and this is just the white and gray matter, and does not include the glycogen body (which is proportionally small in the ostrich, as we’ll see in a future post).

Note that the ostrich does have a much smaller expansion of the spinal cord associated with the forelimbs, but one glance at the graph will tell you that the hindlimbs are a lot more important. This too has implications for fossils. Because the cross-sectional area of the neural canal tends to track the cross-sectional area of the spinal cord (despite the cord not filling the canal), it is possible to make inferences about limb use in fossil taxa based on the relative cross-sectional area of the neural canal along the vertebral column. Emily Giffin published several papers about this in the 1990s (e.g., Giffin 1990, 1995), all of which are worth reading.

Next in this series: the glycogen body.


Dorsal vertebra of a rhea from the LACM ornithology collection. Note the pneumatic foramina in the lateral wall of the neural canal.

If you’ve been here for very long you know I have a bit of a neural canal fixation. Some of this is related to pneumaticity, some of it is related to my interest in the nervous systems of animals, and some of it is pure curiosity about an anatomical region that seems to receive very little attention in proportion to its weirdness – especially in birds.

Human thoracic vertebrae in midsagittal section showing vertebral venous plexus. Gray (1918, image 579), available from

The neural canals of mammals are pretty boring. The canal is occupied by the spinal cord and its supporting layers of meninges, and the rest of the volume is padded out by adipose tissue and blood vessels, notably an extra-dural venous plexus. Aaand that’s about it, as far as I know. (If there are weird things inside mammalian neural canals that I’ve missed, please let me know in the comments – I’m a collector.)

But not so in birds, which have a whole festival of weird stuff going on inside their neural canals. Let’s start with pneumaticity, just to get it out of the way. Many birds have supramedullary diverticula inside their neural canals, and these can leave osteological traces, such as pneumatic foramina, in the walls of the neural canal. That’s cool but it’s a pretty well-known system – see Muller (1908) on the pigeon, Cover (1953) on the turkey, and these previous posts – and I want to get on to other, even stranger things.

The lumbosacral spinal cord of a 3-week-old chick in dorsal view. The big egg-shaped mass in the middle is the glycogen body. Watterson (1949: plate 1).

The spinal cords of birds have several gross morphological specializations not seen in mammals, as do their meninges, and most of these apomorphic structures can also leave diagnostic traces on the inner walls of the neural canal. In fact, birds have so many weird things going on with their spinal cords – at least five different things in the lumbosacral region alone – that I spent a week back in January just sorting them out. To crystalize that body of knowledge while I had it all loaded in RAM, I made a little slideshow for myself, and I’ll use screenshots of those slides to illustrate the morphologies I want to discuss. We’ll cover the vanilla stuff in the next post, and the really weird stuff in subsequent posts.

Stay tuned!


In the first installment in this series (link), we looked at a couple of weird sauropod vertebrae with neurocentral joints that were situated either entirely dorsal or ventral to the neural canals. This post has more examples of what I am calling “offset” neurocentral synchondroses.

I decided it made more sense to refer to the synchondrosis as being offset, instead of referring to the neural canal as offset. Because the neural canal in all of these vertebrae is right where it pretty much always is, just dorsal to the articular surfaces of the centrum. In an adult, fused vertebra, there’d be no sign that anything unusual had ever happened. So I think it makes more sense to talk about the neurocentral joint having migrated dorsally or ventrally relative to the canal, rather than vice versa. If you know differently, or if these weirdos have been addressed before elsewhere and I’ve just missed it, please let me know in the comments!

Here’s a plate from Marsh (1896) showing caudal vertebrae of Camarasaurus (“Morosaurus” in O.C. Marsh parlance), which echo the Alamosaurus caudal from the first post in having the neurocentral joint almost entirely ventral to the neural canal. The neural arch here doesn’t just arch over the canal dorsally, it also cuts under it ventrally, at least in part.

This plate is also nice because it shows the relationships among the arch, centrum, and caudal ribs before they fuse. Here’s the caption, from Marsh (1896):

Here’s the preceding plate, Plate 33, with illustrations of an unfused Camarasaurus sacrum.

And its caption:

This plate not only shows how the sacral ribs fuse to the arch and spine medially, and to each other laterally (forming the sacrocostal yoke), it also shows a last sacral that is very similar to the aforementioned caudals in the position of the neurocentral joint. But interestingly that neurocentral joint offset only seems to be present in the last caudal sacral – the lower figure shows widely-separated neurocentral joint surfaces in the more anterior centra, indicating that the neural arches (not figured in this dorsal view) did not wrap around the neural canal to approach the midline. (I think we’re looking at S2 through S5 here, and missing a dorso-sacral.)

So now I’m freaked out, wondering if this neural arch wrap-around in the caudals is common to most sauropods and I just haven’t looked at enough juvenile caudals to have spotted it before. As always, feel free to ablate my ignorance in the comments, particularly if you know of more published examples. I’m a collector.

The neural canal of that last sacral also has a very interesting cross-sectional shape, like a numeral 8. I have some thoughts on that, but they’ll keep for a future post in this series.

Here’s a dorsal vertebra of Camarasaurus in anterior view (from Ostrom & McIntosh 1966, modified by Wilson & Sereno 1998). It is one of the most disturbing things I have ever seen in a sauropod. It makes my skin crawl.

Here’s why: the centrum and the thing we habitually call the ‘neural arch’ aren’t fully fused, and as this modified version makes clear, the ‘neural arch’ is neither neural nor an arch. Instead of being bounded ventrally by the centrum and dorsally and laterally by the neural arch, the neural canal lies entirely below the synchondrosis between the not-really-an-arch and the centrum.

Why?! WHY WOULD YOU DO THAT, CAMARASAURUS? This is not ‘Nam. This is basic vertebral architecture. There are rules.

Look at c6 of Apatosaurus CM 555 here, behaving as all good vertebrae ought to. Neural arch be archin’, as the kids say.

And if you are seeking solace in the thought that maybe the artist just drew that Cam dorsal incorrectly, forget it. I’ve been to Yale and examined the original specimen. I’ve seen things, man!

Camarasaurus isn’t the only pervert around here. Check this out:

Unfused neural arch of a caudal vertebra of a juvenile Alamosaurus from Big Bend. And I mean, this is a neural arch. This may be the most neural of all neural arches, in that it contains the entire neural canal. It’s more of a neural…ring, I guess. That’s right, this Alamosaurus caudal is batting for the opposite team from the Cam dorsal above. And it’s a team that neither you nor I play on, because we have well-behaved normal-ass vertebrae with neural arches that actually arch, and then stop, like God and Richard Owen intended.

Scientifically, my question about these vertebrae is: well, that is, I mean to say, what!? I think they have damaged me in some fundamental way.

If you have anything more intelligent to add (or even less intelligent – consider the gauntlet thrown down!), the comment thread is open.


  • Ostrom, John H., and John S. McIntosh. 1966. Marsh’s Dinosaurs. Yale University Press, New Haven and London. 388 pages including 65 absurdly beautiful plates.
  • Wilson, J. A. and Paul C. Sereno. 1998. Early evolution and higher-level phylogeny of sauropod dinosaurs. Society of Vertebrate Paleontology, Memoir 5: 1-68.